Stress or strain? Appropriate parameters for predicting the fatigue life of single-crystal nickel-based alloys

This study combines self-designed fatigue tests and literature data to systematically analyze the fatigue fracture mechanisms of nickel-based single-crystal alloys under various temperatures and loading conditions, and their influence on life prediction models. Three traditional models based on maximum shear stress amplitude (Δτ_max), maximum shear strain energy density ((Δτ·Δγ)_max), and strain (or stress) amplitude (Δε/Δσ) were compared across temperatures, alloys, and loading scenarios. Results show that the Δτ_max model is suitable for low-cycle fatigue at low to intermediate temperatures, while the (Δτ·Δγ)_max model better fits high-temperature and complex loading conditions; the Δε/Δσ model performs relatively poorly. Fracture surface analysis reveals a strong temperature dependence of crack propagation mechanisms, supporting the microstructural basis for model selection. To address limitations of traditional models, a three-parameter adaptive model coupling Δτ_max and Δγ_max is proposed. Combined with crystal plasticity computations and a random forest algorithm, a unified fatigue life prediction framework is established. This framework demonstrates stable performance across multiple conditions and significantly outperforms conventional models, showing strong potential for engineering applications.